Maybe like me you are used to read architecture documents that state that an application should have high availability. It’s not uncommon to read that an application should have four-nine availability (99,99%). But what does it really mean ? In the JBoss in Action book there is a very clear explanation by giving the allowed downtime per year. Here is the table :
Allowed downtime per year
3.1 seconds downtime per year, it’s really not a lot. So if you sign a contract with 99.99999% availability, make sure you have good lawyers around you.